Google has been developing a new artificial intelligence program that can create videos from text prompts. The program, called “Phenaki”, was featured on a recent episode of 60 Minutes Overtime, where correspondent Scott Pelley tried it out with some examples.
Phanaki is based on a cascade of video diffusion models, which transform a text prompt into a low-resolution video and then upscale it to high-definition quality. The program can generate videos of various styles and scenarios, such as animations, paintings, 3D objects, and realistic scenes.
Some of the videos created by Phenaki are impressive, uncanny, and unsettling. For instance, Pelley typed in “a teddy bear washing dishes” and got a video of a fuzzy toy scrubbing plates in a kitchen sink. He also typed in “a British shorthair jumping over a couch” and got a video of a cat leaping on a sofa.
However, some of the videos also show limitations and artifacts, such as distorted shapes, blurry textures, and unnatural movements. For example, Pelley typed in “shoveling snow” and got a video of a person with a shovel that looked more like a blob than a human.
Google says that Phenaki is still in a research phase and that it is not releasing it to the public yet. The company acknowledges that there are ethical and safety challenges with such technology, as it could be used for malicious purposes such as creating fake news, propaganda, or pornography.
AI Battle is Moving Across Services
Google also showed a an image generating AI known as MediaGen is similar to other image/video AI's such as MidJourney and Microsoft's Bing Image Creator.
Bing Image Creator is one of Microsoft's latest tools for AI and web services. The tool can create images based on simple words or phrases, such as animals, objects, landscapes, or abstract concepts. However, the tool also has some limitations and restrictions. For example, it refused to make an image based on the word “Bing” when it was first released.
Image Creator launched last month as an image-search accompaniment for Bing Chat. It is powered by OpenAI and Microsoft's DALL-E image processing natural language AI model.
Google is seemingly playing catch up to Microsoft in terms of AI availability and in terms of publicity. Microsoft has already brought AI to the mainstream on Bing Chat, Azure OpenAI Service, and Microsoft 365 Copilot.
Through its partnership with OpenAI and leveraging GPT-4, Microsoft has been grabbing the AI headlines alongside ChatGPT. Although, Google appearing on 60 Minutes certainly shows Mountain View is also ready to play the media game to build its AI profile.
Tip of the day: Windows now has a package manager similar to Linux called “Winget”. In our tutorial, we show you how to install and use this new tool that allows the quick installation of apps via PowerShell or a GUI.
